SEARCH

SEARCH BY CITATION

Keywords:

  • Four and Half Lim Protein;
  • integrins;
  • osteoblasts;
  • differentiation;
  • body composition

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

FHL2, a molecule that interacts with many integrins and transcription factors, was found to play an important role in osteoblast differentiation. Overexpression of FHL2 increases the accumulation of osteoblast differentiation markers and matrix mineralization, whereas FHL2 deficiency results in inhibition of osteoblast differentiation and decreased bone formation.

Introduction: Integrin-matrix interaction plays a critical role in osteoblast function. It has been shown that the cytoplasmic domains of integrin β subunits mediate signal transduction induced by integrin-matrix interaction. We reasoned that the identification of proteins interacting with β-cytoplasmic tails followed by analysis of the function of these proteins would enhance our understanding on integrin signaling and the roles of these proteins in osteoblast activities.

Materials and Methods: Yeast two hybrid assay was used to identify proteins interacting with the cytoplasmic domain of integrin β5 subunit. The association of these proteins with integrin αvβ5 was confirmed by confocal analysis and co-immunoprecipitation. A stable MC3T3-E1 cells line overexpressing Four and Half Lim Protein 2 (FHL2) and mouse osteoblasts deficient in FHL2 were used to study the roles of FHL2 in osteoblast differentiation and bone formation. Matrix protein expression was determined by mRNA analysis and Western blotting. Matrix mineralization was detected by Alizarin red staining. Alkaline phosphatase activity was also measured. μCT was used to determine bone histomorphometry.

Results and Conclusions: FHL2 and actin-binding proteins, palladin and filamin A, were identified as proteins interacting with β5 cytoplasmic domain. FHL2 co-localized with αvβ5 at the focal adhesion sites in association with palladin and filamin A. FHL2 was also present in nuclei. Osteoblasts overexpressing FHL2 exhibited increased adhesion to and migration on matrix proteins. Conversely, FHL2 stimulation of CREB activity was dependent on integrin function because it was inhibited by Gly-Arg-Gly-Asp-Ser (GRGDS) peptide. The expression of osteoblast differentiation markers and Msx2 was upregulated, and bone matrix mineralization was increased in FHL2 overexpressing cells. In contrast, FHL2-deficient bone marrow cells and osteoblasts displayed decreased osteoblast colony formation and differentiation, respectively, compared with wildtype cells. Moreover, FHL2-deficient female mice exhibited greater bone loss than the wildtype littermates after ovariectomy. Thus, FHL2 plays an important role in osteoblast differentiation and bone formation.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

IT HAS BEEN well established that β1- and αv-containing integrins play important roles in osteoblast function because these integrins govern the interaction of osteoblasts with many bone matrix proteins important for bone formation. (1, 2) Osteoblasts express numerous αv-containing integrins. (1, 3, 4) Among these, αvβ5 is one of the most abundant. In fact, this integrin is first purified from osteoblasts. (5) We have previously shown that the maintenance of intact function of αvβ5 is essential in BMP-2 upregulation of alkaline phosphatase (ALP) activity in osteoblasts. (6) However, αvβ5-null mice do not exhibit any obvious bone phenotype, (7) presumably because of the compensatory effects of other function-redundant αvβ integrins present in osteoblasts. Therefore, αvβ5 knockout (KO) osteoblasts may not be suitable for studies on the signaling mechanism involved in αvβ5-matrix interaction. It has been well established that the cytoplasmic domains of β subunits govern signal transduction induced by integrin-matrix interaction. (8) We reasoned that by identifying proteins interacting with the cytoplasmic tail of β5 followed by analysis of the function of these proteins, we might gain insights on the mechanisms mediating αvβ5 activities and the function of these proteins in osteoblasts. Using yeast two hybrid screening, we identified Four and Half Lim Protein (FHL2) and actin-associated proteins, palladin and filamin A, as proteins interacting with αvβ5.

FHL2 belongs to a small family of proteins whose structure is composed entirely of LIM domains (named after the first three proteins, Lin-11, Isl-1, and Mec-3, identified to contain LIM domain). (9, 10) LIM domain is a cysteine- and histidine-rich polypeptide consisting of 50–60 amino acids that can form two adjacent zinc fingers. (11) At present, the LIM domain does not seem to interact directly with DNA but rather participates in protein-protein interaction. (11) FHL2 is a protein of 279 amino acids consisting of a half LIM domain at the N-terminal followed by four full LIM domains. (9, 10) It is present in both cytosol and nuclei. In cytosol, FHL2 can interact with various integrins, including α3, α7, β1, β2, β3, and β6 at the focal adhesion sites. (12, 13) In addition, FHL2 is also shown to bind activated extracellular signal-regulated kinase 2 (ERK2) and prevents its translocation into nuclei in cardiomyocytes. (14) Despite this inhibitory activity, FHL2 is found to translocate from cytosol to nuclei in a Rho GTPase-dependent manner. (15) In the nuclei, FHL2 functions as a co-activator for many transcription factors including activator protein 1 (AP-1), cAMP-responsive element binding protein (CREB), androgen receptor, and β-catenin, which results in modulation of gene expression and cell differentiation. (16–22) Interestingly, FHL2 can also act as a co-repressor in regulating chromatin remodeling by binding to a tissue-specific transcriptional repressor promyelocytic leukemia zinc finger protein (PLZF) and enhancing PLZF-repressing activity. (23) FHL2 not only regulates normal cell function but also is closely associated with cancer cell development. For example, FHL2 can interact with Sloan-Kettering Institute (SKI) oncoprotein in melanoma cells and accelerates melanoma progression, (24) whereas downregulation of FHL2 is detected in rhabdomyosarcoma development. (25) Thus, FHL2 possesses dual functions either as an activator or a repressor depending on the protein partners involved. In osteoblasts, FHL2 is shown to interact with insulin-like growth factor binding protein 5 (IGFBP-5) in nuclei, although the consequence of this interaction has not been explored. (26) Here we report the roles of FHL2 in osteoblasts function and bone formation.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

Yeast two hybrid analysis

Matchmaker GAL4 Two-Hybrid System 3 and U2Os osteosarcoma cDNA library fused to the Gal 4 activation domain in pACT2 plasmid (both from Clontech Laboratories) were used to identify proteins interacting with αvβ5 integrin following the protocols provided by Clontech. In this method, cDNA encoding the cytoplasmic domain of murine β5 (amino acids 719–775) was inserted into the EcoRI/BamHI cloning sites of pGBKT7 plasmid to generate β5-Gal 4 DNA-binding domain fusion protein. This fusion protein was used as the bait, whereas the U2Os cDNA library fused to the Gal 4 activation domain was the prey. Both bait and prey plasmids were co-transfected into yeast AH109 using Yeastmaker Yeast Transformation System 2 (Clontech). AH109 carries four reporter genes (His3, Ade2, LacZ, and Mel1), which are under the control of Gal 4 upstream activating sequences and TATA boxes. When bait and prey interact, the Gal 4 DNA-binding domain and activation domain are brought to proximity, thus activating the transcription of the aforementioned four reporter genes. By repeated growing the transformed yeasts in nutrients lacking histidine and adenine and screening the colonies for the expression of both β- and α-galactosidase activities, plasmids encoding proteins interacting with β5 cytoplasmic domain were isolated using Zymoprep Yeast Plasmid Minipreparation kit (Zymo Research). The cDNA inserts in the plasmids were sequenced and subjected to the Basic Local Alignment Search Tool (BLAST) search to identify proteins interacting with β5 cytoplasmic domain.

Cell cultures

Primary human osteoblasts (HOBs) were isolated from trabecular bone chips of cadaver ribs according to previously published procedures. (27) MC3T3-E1 osteoblastic cell line overexpressing FHL2 was generated by transfection with pcDNA3 plasmid encoding the C-terminal flag-tagged FHL2 followed by G418 selection. All the cells surviving G418 treatment were pooled, and only the first four passaged cells were used for experiments. Cells transfected with pcDNA3 empty vector were used as control. Primary murine osteoblasts were isolated from neonatal calvariae. Cells released in the first 20 minutes of collagenase (2 mg/ml) digestion at 37°C were discarded. The remainders of calvariae were further digested with collagenase and DNase I (5 μg/ml) for 2 h. Enzyme-released cells were harvested, grown to confluence, and subcultured. All the cells were cultured in α-MEM containing 10% FBS.

Immunostaining and confocal analysis

HOBs cultured on cover slips were washed with PBS and fixed in methanol/acetone (1:1, vol/vol). After further PBS wash, cells were stained with primary antibodies followed by Cy3- and Alexa 488-conjugated secondary antibodies as previously described. (6) Fluorescence images were obtained using a BioRad Radiance 2100 Confocal Fluorescence Microscope equipped with LaserSharp Software.

Mammalian two hybrid analysis

Checkmate Mammalian Two-Hybrid System (Promega) was used to confirm the interaction between β5 cytosolic domain and FHL2. In this method, pG5Luc encoding firefly Luciferase cDNA under the control of five consecutive Gal4 binding elements was used as the readout for the interaction between the bait (a pACT plasmid encoding the β5 cytosolic domain and V16 activation domain fusion protein) and prey (a pBIND plasmid encoding the fusion protein of FHL2 and Gal4 DNA binding domain). All three plasmids were co-transfected into ROS 17/2.1 osteosarcoma cells according to the protocol supplied by Promega. For negative control, the pBIND empty vector was used as prey. The interaction between β5 cytosolic domain and FHL2 was expressed as the ratio of firefly luciferase activity over the intrinsic Renilla luciferase activity present in the pBIND plasmid.

Immunoprecipitation and Western blot analysis

Cell layers were washed three times with PBS and extracted with 0.5% Triton X-100 in 10 mM HEPES, pH 7.4, containing 150 mM NaCl and protease inhibitor cocktail. (28) Protein concentration in the extract was measured using a Bio-Rad DC kit. Samples containing equal amount of proteins were subjected to SDS-PAGE directly or were immunoprecipitated with P1F6 anti-αvβ5 antibody (Chemicon International) or control antibody (Santa Cruz). Immune complexes were pulled down using protein A-Sepharose beads. After extensive washing with PBS containing 0.5% Tween 20, bead-proteins were extracted using sample buffer and applied to SDS-PAGE gel. After electrophoresis, proteins were blotted onto an Immobilon P membrane. Membranes were incubated with specific primary antibodies followed by horseradish peroxidase-conjugated secondary antibodies as described previously. (29) Protein bands were visualized by enhanced chemiluminescence using the ECL kit.

Cell adhesion assays

Adhesion to matrix proteins was performed as previously described. (27) Briefly, cells released by sequential collagenase and trypsin/EDTA digestion and resuspended in α-MEM supplemented with 0.1% BSA were seeded (4 × 104/well) in wells that were precoated with BSA (as negative control), vitronectin, fibronectin, or type I collagen (Chemicon). After 1-h incubation at 37°C, nonadherent cells were removed, and the relative number of adherent cells was measured by optical density at 595 nm after fixation and staining with 0.5% toluidine blue in 4% paraformaldehyde and dissolving the stain in 1% SDS.

Cell migration assay

The day before assay, Costar transwell membranes were coated with collagen (100 ng/well) for 3 h at room temperature. Single cell suspensions obtained as described above for adhesion assays were delivered into transwells (100,000 cells in 0.1 ml/transwell) suspended inside the bottom wells containing 0.6 ml of α-MEM with 0.1% BSA and 25 ng/ml FGF2. After overnight migration in an incubator, cells on the membranes were fixed and stained using Leukostat kit (Fisher). Nonmigrated cells on top of the membranes were removed with cold PBS-soaked cotton swabs and the number of migrated cells on the bottom side of the membranes was counted under microscope with the aid of a grid. A total of nine separate fields (3 × 10−4 cm2/field) were counted for each sample, and the average number of cells per field was calculated.

Transfection and luciferase activity assay

MC3T3-E1 stable cell lines overexpressing pcDNA3 or FHL2 were seeded at 150,000/well in 24-well plates the day before transfection. Cells were transfected with pCRE-Luc (Stratagene) using LipofectAmine according to the protocols provided by Invitrogen. Five hours later, cells were treated with 50 μM GRGES or GRGDS peptide in 2% FBS for 16 h. After extraction with reporter lysis buffer (Promega), luciferase activities in cell layers were measured as previously described. (30) To examine the expression of osteocalcin, osteocalcin promoter-Luc (pOC-Luc, −635/+30, kindly provided by Dr Dwight A Towler, WA University, St Louis, MO, USA) was transfected into the stable cell lines and luciferase activity measured 48 h later.

ALP assay

Confluent cell layers were washed with TBS (50 mM Tris, pH 7.4, and 0.15 M NaCl), scraped into 0.5 ml of 50 mM Tris, pH 7.4, and homogenized by sonication. ALP activities in sonicates were determined by measuring the amount of p-nitrophenol released (A405 nm) from p-nitrophenyl phosphate substrate. Enzyme activity was normalized with the protein concentration in sonicate, which was measured by using BioRad protein assay kit.

Mineralization analysis

Cells in 24-well culture plates were grown for 7 days to confluence. Beginning on day 8, cells were treated with α-MEM containing 10% FBS, ascorbic acid (50 μg/ml), and β-glycerophosphate (10 mM) for 3 weeks. Mineral deposition in the matrix was detected by Alizarin red S staining or by the uptake of45Ca as previously described. (29, 31)

RT-PCR and quantitative real-time PCR

Total RNA was purified from cells using RNeasy kit from Quiagen. Two micrograms of total RNA was reverse transcribed (RT) using oligo(dT)15 and random primers and SuperScript II reverse transcriptase (Invitrogen) according to the protocol provided by Invitrogen. The relative level of specific transcript in the first-strand cDNA was analyzed by either semiquantitative PCR or quantitative real-time PCR assays. The primers used for FHL2 in semiquantitative PCR were as follows: forward, 5′-ACTGGAATTCATGACTGAGCGCTTTGACTGC-3′; and reverse, 5′-ACTGGGATCCGGTCTCAAAGCACACCACGCA-3′. PCR was performed using Perkin Elmer Gene Amp PCR System 2400 with the following conditions: 95°C for 3 minutes; 35 cycles of 1 minutes of 95°C, 1 minute of 55°C, and 1 minute of 72°C; 72°C for 15 minutes; and parking at 4°C. Quantitative real-time PCR was performed using ABI PRISM 7000 Sequence Detection System (Applied Biosystems). Gene-specific primers for real-time PCR were designed using the Primer Express software from Applied Biosystems and were listed in Table 1. The relative mRNA level of each protein was expressed as percent of 18S ribosomal RNA level.

Table Table 1.. Sequences of Primers for Quantitative Real-Time PCR
Thumbnail image of

Body composition measurements

Use of mice in this research project was reviewed and approved by the Institutional Animal Care and Use Committee of The Washington University. Body composition was analyzed by DXA using a PIXImus scanner (GE/Lunar). Calibration of the instrument was conducted before each group of measurement as suggested by the manufacturer. Mice anesthetized with ketamin (100 mg/kg) and xylazine (10 mg/kg) intramuscularly were placed on the imaging tray prostrate position. Bone scans were performed and whole body BMD with the exclusion of head was acquired by using the software provided by the manufacturer. To examine the change in BMD after depletion of estrogen, basal BMD was obtained before ovariectomy (OVX) or sham operation (SHAM) in 4-month-old female mice. Subsequent BMD was measured every 2 weeks after surgery, and the percent change in BMD was calculated using the following equation: {(BMD − basal BMD)/basal BMD} × 100.

β-Galactosidase activity staining

Osteoblast cultures and calvariae derived from FHL2-knockout with a concomitant β-galactosidase knockin mice(32) were fixed in 0.5% glutaraldehyde in PBS for 5 minutes. After rinsing with PBS twice (5 minutes each), X-gal (1 mg/ml) in PBS containing 5 mM each of potassium ferricyanide and potassium ferricyanide, and 2 mM MgCl2 was added and incubation continued at 37°C for 3–18 h. Cells containing β-galactosidase were stained blue.

Osteogenesis of bone marrow cells

Bone marrow cells were isolated from 2-month-old mice as described. (33) Cells were seeded in 96-well plate at a density of 100,000 cells/well and left undisturbed for 7 days. Cells were fed with α-MEM medium containing 10% FBS, ascorbic acid (50 μg/ml), and β-glycerophosphate (10 mM) for 3 weeks. Mineralized osteoblastic colonies were detected by Alizarin red S staining. Wells containing mineralized osteoblastic colonies with 15 or more cells were considered positive. The osteogenic potential of bone marrow cells was determined by the percentage of positive wells in all the wells. The percentage of wells containing fibroblastic (including osteoblastic and nonosteoblastic) colonies was also counted to represent the potential of total colony formation.

μCT imaging

The right tibias obtained 1 month after OVX (verified by the uterine size at time of death) or SHAM surgery, which was performed at 4 months of age (n = 4–7/group), were used to assess histomorphometry of trabecular bones and mean cortical thickness using μCT. Specimens were held by embedding in 1.5% agarose gel within plastic vials; the vials were positioned within a 16-mm-diameter acrylic tube. Full-length scans were obtained at an isotropic voxel resolution of 16 μm using a commercial scanner (Scanco μCT 40; SCANCO Medical AG; energy = 55kV, current = 145 mA, integration time = 300 ms). A fixed threshold of 270 was used to separate bone from background. A total of 300–350 transverse CT slices were obtained, and 3D analysis was performed on trabecular bones in the 50 slices starting at about 0.1 mm below the lowest point of the growth plate. Trabecular morphometric parameters including the percentage of bone volume per total volume (%BV/TV), trabecular thickness (Tb Th, μm), trabecular number (mm−1), trabecular spacing (μm), structure model index, and connectivity density (mm−3) were acquired by direct method of calculation rather than one based on stereological models.

To determine the thickness of cortical bones, cortical bone widths in slices 1, 10, 20, 30, 40, and 50 of the aforementioned 50 slices were measured. In each slice, the distances between the outer and inner circumferences of the cortical bones in eight approximately evenly spaced spots were determined and the average thickness of the cortical bone in each slice was obtained. The mean cortical bone thickness (μm) of each bone was calculated from the average cortical thickness of the six slices.

Statistics

Statistical analyses were performed using Student's unpaired t-test. Data were presented as mean ± SE.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

FHL2, palladin, and filamin A co-localize/overlap with integrin αvβ5 at the focal adhesion sites in osteoblasts

Using yeast two hybrid analysis, we identified a C-terminal fragment of palladin (from amino acid 298 to the end of C-terminal, amino acid 789), a C-terminal fragment of filamin A (from amino acid 2439 to the end of C-terminal, amino acid 2647), and a full-length FHL2 as proteins interacting with β5 cytoplasmic domain. Immunostaining for palladin and αvβ5 followed by confocal analysis indicated that αvβ5 adjoined the tip of the palladin fibrils with partial overlapping between these two proteins detectable in some fibrils (Fig. 1A). In contrast, αvβ5 was found to co-localize with filamin A at the end of the filamin A fibrils (Fig. 1B). The association between FHL2 and αvβ5 was also confirmed by immunostaining (Fig. 1C). The interaction between these proteins with αvβ5 was further verified by immunoprecipitation with anti-αvβ5 antibody (P1F6) followed by Western blotting for palladin, filamin A, and FHL2 proteins (Fig. 1D). P1F6 pulled down filamin A, palladin, and FHL2, whereas control antibody did not. Additionally, analysis using CheckMateMammalian Two-Hybrid System in ROS 17/2.8 osteosarcoma cells confirmed the protein-protein interaction between the full-length FHL2 and β5 cytosolic domain (Fig. 1E).

thumbnail image

Figure FIG. 1.. Palladin, filamin A, and FHL2 co-localize/associate with integrin αvβ5. HOB cells cultured on cover slips were immunostained with (A) rabbit anti-palladin and mouse anti-αvβ5 antibodies; (B) mouse anti-filamin A and rabbit anti-αvβ5 antibodies; and (C) rabbit anti-FHL2 and mouse anti-αvβ5 antibodies followed by staining with Cy3-conjugated and Alexa 488-conjugated secondary antibodies. Fluorescence was detected using a BioRad confocal fluorescence microscope. Arrows point to some of the co-localization areas. (D) Palladin, filamin A, and FHL2 co-immunoprecipitate with αvβ5. HOB cell extracts (400 μg) were immunoprecipitated with anti-αvβ5 antibody (P1F6) or control antibody. Immune complexes were pulled down with protein A-Sepharose and subjected to Western blot analysis for palladin, filamin A, and FHL2. (E) Mammalian two hybrid analysis confirmed the interaction between the β5 cytosolic domain and FHL2. ROS 17/2.1 osteosarcoma cells were transfected with pG5Luc, pACT bait containing the β5 cytosolic domain, and pBIND prey alone or with FHL2 insert. Seventy-two hours later, cells were extracted, and luciferase activities were measured. *p < 0.001 vs. pBIND value.20

Download figure to PowerPoint

Because palladin, filamin A, and FHL2 all interacted with β5 cytosolic domain, we examined the spatial relationship between FHL2 and palladin and filamin A. Similar to αvβ5, FHL2 was detected adjoining the end of the palladin fibrils with partial overlapping between these two proteins (Fig. 2A, arrowheads). In addition, FHL2 was also found to be present along some of the palladin fibrils (Fig. 2A, arrows). Like αvβ5, FHL2 co-localized/overlapped with filamin A at the tips of the filamin A fibrils (Fig. 2B, arrowheads). In addition, FHL2 was also found to be present along some of the filamin A fibrils (Fig. 2B, arrows).

thumbnail image

Figure FIG. 2.. FHL2 co-localizes/overlaps with palladin and filamin A. HOB cells on coverslips were immunostained with rabbit anti-FHL2 and (A) mouse anti-palladin or (B) mouse anti-filamin A antibodies followed by staining with Cy3-conjugated anti-rabbit IgG and Alexa 488-conjugated anti-mouse IgG. Fluorescence was detected using a BioRad confocal fluorescence microscope. Arrowheads point to the co-localization/overlapping of FHL2 with palladin or filamin A. Arrows point to the presence of FHL2 on some of the palladin and filamin A fibrils.20

Download figure to PowerPoint

Because both palladin and filamin A were actin-associated proteins, (34–36) they were most likely involved in linking αvβ5 at the focal adhesion sites to cytoskeletons. FHL2 has been shown to bind to a variety of integrins including αβ1 and αvβ3(12, 13) in addition to the αvβ5 reported here. FHL2 has also been shown to shuttle between cytosol and nuclei. (15) Therefore, FHL2 has the potential to act as a common messenger in transmitting signals induced by integrin-matrix interaction from the focal adhesion sites into nuclei to regulate gene expression. We chose to study in detail the roles of FHL2 in osteoblast differentiation and bone formation.

FHL2 and integrins regulate osteoblast activities interdependently

Because integrins played important roles in cell adhesion and migration, (27, 37, 38) we used these properties to examine the functional relationship between integrins and FHL2. To this end, we generated a FHL2 overexpressing MC3T3-E1 stable osteoblastic cell line using a pcDNA3 plasmid encoding the C-terminal Flag-tagged FHL2. Stable cell line obtained using pcDNA3 empty vector was the control. RT-PCR analysis and Western blotting indicated that FHL2 levels were 2- to 3-fold of control levels in FHL2 overexpressing cell line (Figs. 3A and 3B). More importantly, the Flag-tagged FHL2 was found to be expressed at the focal adhesion sites (Fig. 3C, arrowheads) and in the nuclei when cells were stained with anti-Flag antibody (Fig. 3C, arrow).

thumbnail image

Figure FIG. 3.. FHL2 enhances osteoblast migration and adhesion and stimulates pCRE-Luc expression. (A) Semiquantitative RT-PCR showed that FHL2 was overexpressed in FHL2 cell line. (B) Western blot analysis for FHL2 protein and the Flag-tag confirmed the overexpression of FHL2. The levels of tubulin were used as loading controls. (C) FHL2 was expressed at the focal adhesion sites (arrow heads) and in the nuclei (arrow). MC3T3-E1 stable cell line overexpressing Flag-tagged FHL2 was immunostained with mouse anti-Flag antibody. Immune complexes were detected using Cy3-conjugated anti-mouse antibody. (D) pcDNA3 and FHL2 cell lines were subjected to migration on type I collagen-coated transwells. FGF2 (25 ng/ml) delivered to the bottom chamber was used as chemoattractant. After 24-h migration, cell numbers at the bottom side of the transwell membranes were counted. *p < 0.001. (E) pcDNA3 and FHL2 cell lines were subjected to adhesion to vitronectin, fibronectin, or type I collagen for 1 h. The relative number of adherent cells was determined by optical density at 595 nm after fixation and staining with 0.5% toluidine blue in 4% paraformaldehyde and dissolving the stain in 1% SDS. *p < 0.001; **p < 0.01 vs. pcDNA3 values. (F) pcDNA3 and FHL2 expressing stable MC3T3-E1 cell lines were transfected with pCRE-Luc. Luciferase activity was measured 24 h later. *p < 0.001. (G) pcDNA3 and FHL2 expressing stable MC3T3-E1 cell lines were transfected with pCRE-Luc. Five hours later, cells were treated with either GRGES or GRGDS peptide (50 μM) in 2% FBS. Luciferase activity was measured 16 h later.ap < 0.05 vs. the corresponding pcDNA3 value;bp < 0.05 vs. the FHL2 value in the presence of GRGES.20

Download figure to PowerPoint

Examination of cell migration on type I collagen indicated that FHL2 overexpression enhanced osteoblast migration to 2.2-fold of control levels by FHL2 (Fig. 3D). FHL2 also increased osteoblast adhesion to vitronectin, fibronectin, and type I collagen to 2-, 1.8-, and 1.6-fold, respectively, of the control levels (Fig. 3E). These data indicate that FHL2 can enhance integrin activity in cell adhesion and migration. Because it is well known that osteoblast precursor cells residing in bone marrow need to migrate and adhere to bone surface before bone formation can occur, the stimulation of integrin activity by FHL2 may have beneficial effect on bone formation.

FHL2 has been shown to bind and upregulate CREB activity. (16) We took advantage of this property to examine the role of integrins in FHL2 function. As expected, FHL2 overexpression enhanced pCRE-Luc activity to 2.5-fold of the pcDNA3 level (Fig. 3F). This upregulation, however, was abolished in the presence of GRGDS peptide, which inhibits the function of many integrins(39) (Fig. 3G, compare between the third and fourth columns). In contrast, FHL2 maintained the upregulation of pCRE-Luc activity in the presence of control peptide GRGES (1.9-fold of pcDNA3 level, Fig. 3G, compare between the first and second columns). Thus, the upregulation of CREB activity by FHL2 is dependent on integrin function.

Overexpression of FHL2 in osteoblasts stimulates cell proliferation and differentiation

Amaar et al. (26) previously showed that FHL2 interacted with IGFBP-5 in osteoblasts and suggested that FHL2 might function as a co-activator of IGFBP-5 in stimulation of osteoblast proliferation. We found that the proliferation rate of FHL2 overexpressing osteoblasts was indeed increased because the total cell number after a 7-day culture period was higher than that of the control cells (131.8 ± 13.0% of control, p < 0.05). In addition, FHL2 overexpression also enhanced osteoblast differentiation as shown by increased ALP activity (2.1-fold of pcDNA3 cell level; Fig. 4A), osteocalcin (OC) promoter activity (1.8-fold; Fig. 4B), and the expression of bone sialoprotein (BSP) and osteopontin (OPN) proteins (2.6- and 2.8-fold, respectively, of the pcDNA3 levels; Fig. 4C). Examination of the relative mRNA levels as determined by RT followed by real-time PCR confirmed the upregulation of OC, BSP, and OPN in FHL2 cells (Fig. 4D; 3.32 ± 0.46, 2.12 ± 0.33, and 1.53 ± 0.05 fold, respectively). Interestingly, the mRNA level of osterix (Osx), an osteoblast-specific transcription factor essential for differentiation, was increased to 1.98 ± 0.02-fold of control level, whereas Runx2 mRNA was not altered in FHL2 overexpressing cells (Fig. 4D). In addition, mRNA level of Msx2, a homeobox transcription factor important for osteogenesis, (40) in FHL2 overexpressing MC3T3-E1 cells was increased to 3.45 ± 0.09-fold of the pcDNA3 control cell level (Fig. 4D). Consistently, matrix mineralization in FHL2 overexpressing cells was enhanced as shown by Alizarin red S staining (Fig. 4E) and increased45Ca uptake (148 ± 16% of control cells; Fig. 4F). Thus, FHL2 stimulates osteoblast differentiation by increasing ALP activity and the expression of bone matrix proteins and enhancing matrix mineralization.

thumbnail image

Figure FIG. 4.. FHL2 enhances osteoblast differentiation. (A) FHL2 stimulated ALP activity. pcDNA3 and FHL2 cells were seeded in 24-well plates. After 14 days, ALP activities in cell layers were measured. (B) FHL2 stimulated osteocalcin expression. pcDNA3 and FHL2 cells were transfected with pOC-Luc, and luciferase activities were measured 48 h later. (C) FHL2 increased the expression of BSP and OPN. pcDNA3 and FHL2 cells were cultured for 2 weeks. Cell extracts were subjected to Western blot analysis and probed for BSP, OPN, and tubulin. (D) pcDNA3 or FHL2 overexpressing MC3T3-E1 cells were cultured for 2 weeks and analyzed for the relative mRNA levels of OC, BSP, OPN, Osx, Runx2, and Msx2 by quantitative real-time PCR. (E and F) FHL2 increased matrix mineralization. pcDNA3 and FHL2 cells were seeded and allowed to grow for 7 days. Cells were treated with ascorbic acid (50 μg/ml) and 10 mM β-glycerophosphate for 3 weeks. Mineralized matrix was detected by (E) Alizarin red S staining or by (F) measuring the incorporation of45Ca into cell layers. *p < 0.001; **p < 0.01; ***p < 0.05 vs. pcDNA3 values.20

Download figure to PowerPoint

FHL2 deficiency leads to decreased osteoblast differentiation and accelerated bone loss after OVX

To further examine the role of FHL2 in bone formation, we used FHL2-deficient (KO) mice in which the FHL2 gene was replaced with a LacZ cDNA. (32) β-Galactosidase activity staining indicated that FHL2 was highly expressed in neonatal calvaria (blue stain), especially in the suture regions and in osteoblasts lining the bone surface (Figs. 5A and 5B, respectively). Osteoblasts derived from KO calvariae were also stained positive for β-galactosidase activity, whereas those from the wildtype (WT) calvariae were negative (Fig. 5C). Quantitative real-time PCR confirmed the absence of FHL2 in osteoblasts isolated from KO calvariae (Fig. 5D). In addition to FHL2, osteoblasts also expressed significant amount of FHL1, whereas very little FHL3 and FHL4 were detected (Fig. 5D). Bone marrow stromal cells (BMSCs), which served as the source of osteoblast precursor cells, also expressed FHL2 to approximately the same extent as osteoblasts (data not shown).

thumbnail image

Figure FIG. 5.. FHL2 is expressed in calvaria and osteoblasts. (A) β-galactosidase activity (FHL2) was present at high level in bones and sutures of neonatal calvaria isolated from FHL2 knockout (KO) with a concomitant β-galactosidase knockin mice (right). No β-galactosidase activity was detected in the calvaria isolated from WT mice (left). (B) β-galactosidase activity was detected in osteoblasts lining KO calvaria. (C) Osteoblasts isolated from FHL2 KO calvariae also exhibited β-galactosidase activity, whereas wildtype (WT) cells did not. (D) Osteoblasts expressed abundant FHL1 and FHL2 but very little FHL3 and FHL4 mRNA as determined by RT followed by quantitative real-time PCR.20

Download figure to PowerPoint

To determine the osteogenic potential of osteoblastic lineage cells, we first used bone marrow cells (BMCs) that contained BMSCs. BMC derived from FHL2 KO mice developed fewer matrix-mineralizing osteoblastic colonies in cultures than the WT cells (45.8 ± 5.3% versus 84.0 ± 0.7%, KO versus WT; Fig. 6A, compare between the left two columns). Analysis of total fibroblastic colony formation indicated that KO BMCs again had lower potential than the WT cells (59.7 ± 3.0% versus 89.6 ± 1.2%, KO versus WT; Fig. 6A, compare between the right two columns). Because the decline in fibroblastic colony formation was less than that of osteoblastic colony formation in KO BMCs, it appeared likely that FHL2 deficiency not only diminished the stromal cell pool in bone marrow but also decreased their osteogenic differentiation.

thumbnail image

Figure FIG. 6.. FHL2-deficient osteoblastic cells exhibit decreased differentiation. (A) FHL2-deficient bone marrow cells generated fewer total fibroblastic (Total) and osteoblastic (OB) colonies compared with WT cells. Limited diluted bone marrow cells were seeded in 96-well plates for 1 week followed by treatment with ascorbic acid (50 μg/ml) and β-glycerophosphate (10 mM) for 3 weeks. Percentage of wells containing fibroblastic or Alizarin red S-positive colonies with cell number >15 was determined.ap < 0.05 vs. WT;bp < 0.05 vs. the KO Total value. (B and C) Osteoblasts isolated from WT or FHL2 KO calvariae were cultured for 2 weeks and analyzed for (B) the relative mRNA levels of ALP, OC, BSP, OPN, Osx, Runx2, and Msx2 by quantitative real-time PCR and (C) ALP activity. *p < 0.001; **p < 0.01; ***p < 0.05 vs. WT values. (D) Osteoblast cultures were treated with ascorbic acid and βglycerophosphate for 18 days. Matrix mineralization was analyzed by Alizarin red S staining.20

Download figure to PowerPoint

To further explore the possibility that the differentiation of osteoblasts was reduced in the absence of FHL2, we used FHL2 KO calvarial osteoblasts and analyzed their expression of differentiation gene markers. As shown in Fig. 6B, the expression of ALP, OC, BSP, OPN, Osx, and Msx2 was decreased by FHL2 deficiency, whereas Runx2 mRNA level was not altered. Consistently, ALP activity and matrix mineralization of FHL2 KO osteoblasts were suppressed compared with WT osteoblasts (Figs. 6C and 6D, respectively). Thus, FHL2 deficiency leads to decreased osteoblast differentiation in vitro.

In the in vivo setting, temporal body composition examination of female mice indicated, however, that no significant difference in BMD was detected between WT and FHL2 KO mice during the first 4 months of age (Fig. 7A). Nevertheless, KO mice exhibited an accelerated bone loss after OVX. The BMD of FHL2 KO mice was decreased by 4.0% and 4.8% in 2 and 4 weeks, respectively, after OVX, whereas WT littermates reduced BMD by only 1.3% and 0.2%, respectively, during the same time frame (Fig. 7B). The sham-operated WT and KO mice continued to gain BMD throughout the experimental period (Fig. 7B).

thumbnail image

Figure FIG. 7.. FHL2 deficiency leads to decreased bone formation after OVX. (A) FHL2 KO (n = 11) female mice did not show significant changes in BMD compared with WT littermates (n = 9) under basal condition at ages of 1–4 months. (B) OVX KO mice had greater bone loss compared with the WT littermates. The basal BMD of both WT and KO female mice were measured before OVX or sham operation. After operation, BMD of each mouse was measured every 2 weeks. The percent change in BMD from the basal level was calculated. (C-F) μCT analysis of right tibias obtained 1 month after OVX indicated that FHL2 KO had decreased (C) %BV/TV, (D) trabecular thickness, and (E) connectivity density, but (F) an increased structure model index. No differences were detected in these parameters between WT and KO sham-operated tibias. *p < 0.001; ***p < 0.05 vs. the corresponding WT values.20

Download figure to PowerPoint

Consistently, analysis using μCT on the trabecular bones of tibias isolated 1 month after OVX or sham operation indicated that KO OVX mice had decreased bone volume (Fig. 7C; 48.4 ± 0.02% of WT OVX levels), trabecular thickness (Fig. 7D; 70.6 ± 0.02% of WT OVX levels), and connectivity density (Fig. 7E; 50.7 ± 0.1% of WT OVX levels) compared with WT OVX mice. In contrast, the structure model index in KO OVX mice was increased (Fig. 7F; 131.3 ± 0.04% of WT OVX levels), suggesting that the trabecular bones in KO OVX tibias were becoming more rod-like rather than plate-like. No significant differences were detected in trabecular number (2.81 ± 0.30 versus 2.65 ± 0.08/mm, WT OVX versus KO OVX) and trabecular spacing (385 ± 34 versus 382 ± 13 μm, WT OVX versus KO OVX) between WT OVX and KO OVX tibias. Among sham-operated groups, no differences in all the parameters were detected between WT and KO tibias (Figs. 7C–7F). Examination of the mean cortical bone thickness in tibias indicated that FHL2 deficiency resulted in thinner cortical bones compared with WT after OVX (148.8 ± 2.5 versus 138.7 ± 2.0 μm, WT versus KO, p < 0.05, ∼7% decrease), whereas no significant difference in the cortical bone thickness was detected between the sham-operated WT and KO mice (155.4 ± 1.8 versus 153.7 ± 5.3 μm, WT versus KO). These data indicate that FHL2 plays an important role in bone formation under high bone turnover condition induced by OVX. Furthermore, the decrease in BMD in KO OVX mice is derived mainly from reduction in trabecular bone volume and cortical thickness.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

This study was initiated to gain insight on the signaling molecules involved in integrin functions in osteoblasts. We have identified FHL2 as one of the proteins interacting with the β5 cytosolic domain. Because FHL2 can interact with many members of the integrin family, including αβ1, αvβ3, αvβ6, and αvβ5(12) (and this report), it seems likely that FHL2 is a common molecule involved in signal transduction induced by various integrin-matrix interactions. Whereas FHL2 can enhance integrin function, FHL2 activity is, at least in part, dependent on integrin function. The expression of FHL2 was originally thought to be limited primarily to heart, aorta, skeletal muscles, ovary, placenta, adrenal gland, and pituitary gland. (9, 10, 16, 25, 41) Recently, FHL2 was also found to be abundantly expressed in osteoblasts. (26) Our preliminary data suggest that the expression of FHL2 mRNA in osteoblasts is ∼2.5- to 4.7-fold of aorta levels (data not shown).

The role of FHL2 in osteoblasts is virtually unknown despite the demonstration that FHL2 is associated with IGFBP5, which is known to stimulate osteoblast proliferation and differentiation. (26, 42) We have shown that FHL2 is important in bone formation both in vitro and in vivo. Osteoblasts overexpressing FHL2 have increased proliferation and differentiation, whereas FHL2 deficiency leads to decreased osteoblast differentiation and accelerated bone loss after OVX. It is true that accelerated differentiation can be, at least in part, a consequence of increased proliferation because the cell layer becomes confluent faster and can assume differentiation earlier. However, the ability of FHL2 to upregulate OC promoter activity in transfection experiment using nonproliferating confluent cells (Fig. 4B) suggests that FHL2 can indeed stimulate osteoblast differentiation without the benefit of cell proliferation.

It has been shown that different LIM domains govern FHL2 interaction with different proteins. (12, 18, 21, 26, 43) We have performed the yeast two hybrid screening using preys consisting of various FHL2 mutants that prevent zinc finger formation(44) or FHL2 deletion constructs. Mutations in each full LIM domain reduced the potential of FHL2 to interact with β5 to ∼50-65% of the native FHL2 level (data not shown), whereas all the FHL2 deletion constructs encoding various combinations of the one-half and full LIM domains were only capable of interaction with the β5 cytosolic domain at <20% of the full length FHL2 level (data not shown). Thus, the whole molecule of FHL2 is required in achieving optimal interaction with β5.

FHL2 is known to be present in both cytosol and nuclei. In cytosol, FHL2 is most notably co-localized with integrins at the focal adhesion sites and can augment integrin activity(10, 12, 13, 44, 45) (and this report). In nuclei, FHL2 is known to function as a co-activator for several transcription factors including CREB, AP-1, androgen receptor, and β-catenin. (16–22) Because these transcription factors and integrins play critical roles in osteoblast function and bone formation, (2, 6, 46–51) the enhancement in activities of these transcription factors and integrins by FHL2 may explain, at least in part, the osteogenic effect of FHL2. The stimulation of osteoblast specific transcription factor Osx lends additional support to the osteogenic effect of FHL2. Although Osx has been shown to be a downstream effecter of Runx2, (52) we did not detect any alteration in Runx2 mRNA levels by FHL2. The disparity in the regulation of Osx and Runx2 expression has also been reported previously, (40, 53) suggesting that Osx may also be regulated in a Runx2-independent manner. The detailed mechanism in the regulation of Osx by FHL2 awaits future analysis.

In addition to promoting the activities of the aforementioned transcription factors, FHL2 also regulates the expression of Msx2. Overexpression of FHL2 led to increased Msx2 mRNA levels, whereas deficiency in FHL2 resulted in suppression of Msx2. We previously showed that Msx2 can promote osteoblast lineage allocation with a concomitant suppression of pluripotent mesenchymal cell adipogenesis. (40) Consistent with their respective Msx2 levels, osteoblast differentiation and bone formation were suppressed in FHL2-deficient mice, whereas osteoblast differentiation was upregulated in MC3T3-E1 osteoblastic cells overexpressing FHL2. Moreover, we have observed an increased adiposity in FHL2 KO mice (17.59 ± 1.04% versus 20.03 ± 0.83% body fat at 4 months of age, WT versus KO, p < 0.05), consistent with decreased Msx2 levels. These data suggest that Msx2 may play an important role in mediating part of FHL2 activities in body composition regulation. Although the detailed mechanism mediating the regulation of Msx2 expression by FHL2 awaits future elucidation, the potentiation of β-catenin activity by FHL2 and the stimulation of Msx2 expression by β-catenin(18, 21, 22, 54) suggest that β-catenin may play a critical role in FHL2 regulation of Msx2 expression.

One of the most notable functions of FHL2 is its ability to shuttle between cytosol and nuclei. Because FHL2 co-localizes with various integrins at the focal adhesion sites(12, 13) (and this report), this spatial arrangement enables FHL2 to sense integrin-matrix interaction in situ and can function as a messenger to transmit signals into nuclei either by itself or as a carrier for various signaling molecules through protein-protein interaction through the LIM domains. It has been shown that the translocation of FHL2 into nuclei is dependent on Rho GTPase but not Rac or Cdc42. (15) Rho GTPase is known to stimulate the bundling of actin filaments into stress fibers in addition to gene regulation. (55, 56) We have shown that FHL2 is present not only at the end of palladin and filamin A fibrils but also along some of these fibrils. Because palladin and filamin A are localized to actin stress fibers, (34–36) they may participate in Rho GTPase-mediated trafficking of FHL2 between focal adhesion sites and nuclei. Future analysis using knockdown or knockout techniques(34) should provide more definitive answers on the roles of actin stress fibers and associated filamin A and palladin in FHL2 trafficking and function.

Both FHL1 and FHL2 are highly expressed in osteoblasts. The expression of FHL1 in osteoblasts has been reported previously. (57, 58) The marginal bone phenotype observed in FHL2-deficient mice under basal conditions suggests that the functional redundant FHL1 may compensate for the lack of FHL2. However, this compensation does not seem to be sufficient under stressful conditions. The accelerated bone loss after OVX in FHL2-deficient mice compared with WT littermates indicates that FHL2 is important for optimal osteoblast function and bone formation.

While this manuscript was being revised, Gunther et al. (59) published a paper showing that FHL2-deficient mice have osteopenia under basal condition, which is derived primarily from decreased osteoblast activity because the numbers of osteoblasts and osteoclasts are not altered. The discrepancy between our observation and their report may derive primarily from the differences in genetic backgrounds. The genetic background of our FHL2 deficiency mice is in Swiss black/129-SV/J, whereas that of the mice of Gunther et al. is in C57BL/6. It is well established that differences in genetic background contribute to disparity in bone parameters and variation in skeletal responses to manipulations. (60–63) Regardless, the fact that osteopenia is detected in FHL2-deficient mice with the C57/BL/6 background under basal conditions further supports the importance of FHL2 in bone formation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

We thank Dr Dwight A Towler for assistance in preparation of this manuscript. This work was supported by National Institutes of Health Grants AR32087 (SLC), AR07033 (CFL), and HL66100 (JC). PHC was supported by grants from NHRI-EX91-9108SC and NHRI-EX92-9108SC.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References